This paper proposes a digital twin approach that utilizes multiple camera sensors to construct a virtual environment synchronized with the real environment. A 3D reconstruction model of the real environment was generated using multiple camera sensors and the GPU-Voxels package. An efficient transformation process was proposed to make the generated data available for use in the Isaac SIM platform. It was confirmed that a virtual environment synchronized with the real environment was constructed in the Isaac SIM platform using the transformed data. Additional experiments verified that robot navigation tasks worked well in the constructed virtual environment.